Integrating Supervised and Probabilistic Learning: A Hybrid SVM-Naive Bayes Framework for Enhanced Emotional Insight Extraction from Twitter Streams
Mary Harin Fernandez F, Thaduri Venkata Ramana, Srikanth Lakumarapu · 2024
Online social networks have become essential platforms for individuals to share their thoughts and opinions with friends, family, and the broader community on various topics. Through text, picture, audio, and video communications, people may communicate their ideas, feelings, and opinions on social, political, and other issues. Text is still the most popular way for people to express themselves on social networks, even with the wide range of communication types accessible. The aim of this research is to identify and examine the emotions expressed by students via their Twitter tweets. The COVID-19 pandemic has affected students in a variety of ways, resulting in a rise in social media posts highlighting these problems, lockdowns, and related hashtags, both positively and negatively. The study assesses public opinion on actions taken during such times, as well as the different emotional reactions of students to these stressors and related issues. This is accomplished by applying a combination of probabilistic learning and supervised learning using SVM and Naive Bayes classifiers for text categorization, offering a rapid and accurate approach. Predictive models are trained using the Semantic-based Feature Extraction and Feature Selection techniques. This approach allows for the detection of a wide variety of expressed emotions, making it easier to provide appropriate support when necessary.